Pre-Employment Testing for Small Businesses in Australia & New Zealand

The short answer

Pre-employment testing for a small business is a short online assessment — typically 30–45 minutes covering role-relevant aptitude, workplace judgement and, where the role is customer-facing, working English — that every applicant takes under the same conditions before anyone gets an interview. Applicants open a share link or QR code on any phone, create no account, and are scored and ranked automatically the moment they submit. You interview the three people the measurement says are worth an hour, instead of the eight whose applications read best.

The reason this matters more for a small business than a large one is arithmetic: a bad hire in a 400-person company is a rounding error, while a bad hire in an 8-person business is an eighth of the workforce. Yet most assessment platforms are priced and packaged for the 400-person company — annual licences, per-seat pricing, procurement calls — which is why so many Australian and New Zealand SMEs still hire on CV and gut feel. Pay-per-candidate pricing removes that barrier: testing a whole applicant pool costs less than one recruitment-agency placement fee.

Why small businesses skip testing — and why the maths says not to

The usual objections are cost, time and formality. Cost, because incumbent platforms quote annual subscriptions that make no sense for a business hiring five people a year. Time, because the owner is the HR department and cannot spend a week setting up a testing programme. Formality, because structured assessment feels like something only corporates do.

Each objection was fair once and is now out of date. Metered platforms price per candidate assessed — single dollars, not thousands a year — so a business that hires twice a year pays only for those two screens. Setup is composing an assessment from a plain-language description of the role and sharing one link; the first screen can be live the same afternoon. And the formality is exactly the point: hiring on unstructured interviews alone is close to hiring at random, because interviews reward confidence and rapport, which are not the same thing as capability. A short structured screen is the single cheapest upgrade an SME can make to hiring accuracy.

The cost of skipping it is concrete. A mis-hire costs two to three months of wages plus the re-hiring cycle plus, in a small team, the customers and morale that one wrong person can burn through. Against that, screening an entire 30-applicant pool costs roughly what one of those applicants earns in a day.

What to test, role by role

Retail and hospitality: customer-scenario judgement first — what the applicant does with an unreasonable complaint, a queue at close, a till discrepancy — plus attention to detail and basic numeracy. Roster reliability behaviours (turning up, flagging problems early, handling a shift swap honestly) are measurable with situational judgement scenarios and predict more than any CV line.

Office, admin and bookkeeping-adjacent roles: attention to detail and numerical accuracy are the core signal, with written English where the role fronts email. These are the roles where a polished application most often outruns actual accuracy — and where a 20-minute measured screen most often reverses the shortlist.

Care-adjacent and community-facing roles: situational judgement around empathy, boundaries, escalation and following procedure. An assessment measures capability and judgement only — it does not replace the registrations, licences, police or working-with-children checks a role legally requires. Run both; they answer different questions.

Trades-adjacent hiring (apprentices, trade assistants, scheduling and front-office staff for a trades business): mechanical and spatial reasoning for the tools side, numeracy for quoting and materials, and reliability judgement for everyone — because in a three-van business, the person who does not show up is a third of your capacity.

Across every role, keep the test job-related: measure what the job actually demands, at the level it demands it, for every applicant equally. That is both what makes a screen predictive and what makes it fair.

The AI problem: why you cannot just email a take-home task

The informal SME screen — email the applicant a task, see what comes back — stopped working when capable AI assistants became universal. A written exercise done unsupervised now measures who has the better chatbot, not who is the better candidate. The same applies to cover letters and, increasingly, to polished CVs themselves.

Proctored assessment closes the gap without any extra work on your side: an identity photo is captured at the start of the sitting, the camera monitors face presence throughout, tab switches and screen-capture attempts are recorded, and every attempt receives a High/Medium/Low integrity rating with the underlying evidence attached for you to review. Spoken-English tasks are scored from the candidate's own recorded voice, which no chatbot can sit. For a small business this is the difference between trusting a result and hoping about it.

Running a screen without an HR department

Step 1 — compose the assessment. Describe the role in plain language — "front-of-house for a 40-seat café in Wellington, weekend shifts, handles complaints and the till" — and AI composes a job-relevant assessment: the scenarios, the aptitude modules, the scoring. Ready-made aptitude and English modules exist where composition is unnecessary.

Step 2 — share one link. Every job ad, from Seek to a printed sign in the window, carries the same link or QR code. Applicants take the assessment on their own phone at their own time, with no account to create; attempts are resumable, so a dropped connection pauses a sitting rather than ending it.

Step 3 — interview from the ranking. Every submission is scored on arrival: per-section scores, the integrity rating with its evidence, an AI-written summary of strengths and risks, and a live ranking of the whole pool. The owner's job shrinks to reviewing a ranked list and interviewing the top few — the part a human should actually do.

What it costs

Pricing is per candidate assessed, with no platform fee, seat licences or annual contract: credits cost US$0.50 each and a typical screening assessment set runs 15–26 credits — about US$7.50–13 per candidate — with credit packs (250 / 1,000 / 5,000 at US$125 / 465 / 2,175) discounting volume by up to 13%. Prices are quoted and charged in US dollars; your card provider converts to Australian or New Zealand dollars at its prevailing rate. New organisations get 250 free credits — enough to screen an entire applicant pool for one role before paying anything.

Compare the alternatives an Australian SME is usually quoted: assessment-platform subscriptions that only amortise at continuous hiring volume, recruitment-agency fees in the thousands per placement, or consultancy-run psychometrics priced per report at many times the metered rate. Metered testing inverts the economics — the less often you hire, the more per-candidate pricing wins.

Two questions to ask before you sign, added September 2026

Ask for the denominator, not just the price. A monthly figure is not a cost per candidate until you know how many candidates it buys, and that number is published far less often than the price is. Checked at each provider's own page on 9 September 2026, two of six providers an Australian or New Zealand employer commonly shortlists publish a price at all — and of those, one also publishes what the price covers: recruitment plans quoted at 50, 100 and 200 AI video interviews, with the 50-interview plan stated at A$2,388 a year on annual billing, all prices in Australian dollars excluding GST. That divides to A$47.76 per interview if you use all fifty and A$119.40 each if you use twenty. Under-use is the expensive failure mode of an allowance plan and it never shows up in a demo, because a demo is always run at full use. The full six-option table is in Assessment platforms compared: Australia & New Zealand. Criteria, which now owns the Australian brand Revelian and publishes no price, is compared row by row in Criteria Corp alternatives for Australia and New Zealand.

Ask what your privacy policy will have to say from 10 December 2026. The OAIC states that from that date an Australian APP entity's privacy policy must carry information about the kinds of personal information used in automated decision-making and the kinds of decisions made using it, where a decision could reasonably be expected to significantly affect an individual's rights or interests — a change made by the Privacy and Other Legislation Amendment Act 2024. An automated hiring sift is the textbook case. Notice what it is not: it does not ban automated screening, it does not add a consent step, and it does not require you to publish your cut score. It requires you to describe, in your own privacy policy, what the software decides and what information it uses.

The reason this belongs in procurement rather than in a compliance file is that the obligation lands on your privacy policy, not your vendor's — buying a platform does not discharge it — and you cannot draft it without two sentences only the vendor can supply: what kinds of personal information their program uses, and what kinds of decisions it makes or substantially and directly contributes to. Ask for both in writing now. A supplier who cannot produce them in September will not be easier to reach in December. This is a plain-language summary rather than legal advice; take Australian advice on your own facts.

New Zealand employers have a different question to settle first. Under IPP 12 of the Privacy Act 2020, an agency may disclose personal information to a foreign person or entity only if one of six grounds is met — including that the recipient is subject to privacy laws providing comparable safeguards overall, or that the agency otherwise reasonably believes comparable protections apply, which in practice means by contract. Consent is one of the six and is the weakest in a hiring context, because the applicant is authorising the organisation deciding whether to employ them. So the data-processing terms are part of the purchase decision. AssessAll publishes no Australian or New Zealand data-residency option and holds no local accreditation of any kind.

For RTOs and training providers: measure before you train

The same platform serves the other side of the ANZ skills economy: registered training organisations, private providers and workplace-training consultancies who need evidence of where a cohort actually starts. TNA Studio composes a bespoke training-needs analysis from a plain-language brief — the cohort, the context, the capability question — and delivers it through the same share-link, no-accounts flow. Every participant receives an individual report; the programme designer receives a group report that ranks the cohort's gaps by consequence and separates the gaps training will close from the ones it will not.

Learning Journeys then re-measure the same competencies at baseline, during and after a programme, turning "the training went well" into a documented capability delta — evidence for an employer client, a funding conversation or a renewal. For a provider selling into SMEs, a measured baseline is also the most honest sales instrument there is: it scopes the training to gaps that are real. Capability measurement only — AssessAll makes no claims regarding RTO regulatory standards or accredited-training requirements.

Who published a price in September 2026, and who published what it buys (added 16 September 2026)

This site has published a price-transparency check for three markets, and the standing rule is that a comparison row is re-verified monthly or deleted. Re-read at source on 16 September 2026, with a harder question attached than the one asked in the earlier rounds.

TestGorilla still publishes plan prices. Free at US$0 a month with ten free credits a month; Core at US$142 a month, US$1,704 billed annually; Plus starting from US$400 a month, from US$4,800 annually; Enterprise on request. It also publishes a credit rate — one credit per candidate for a skill test or an invitation, two for an AI interview.

Mercer Mettl's pricing page carries no number in any currency. What it carries is a line offering a customised plan and a telephone number. That is a normal enterprise model and it is not a criticism; it does mean every figure circulating for Mettl comes from third-party procurement aggregators rather than from Mettl, and a small business should cite none of them.

The harder question is the one this site raised for Australia and New Zealand on 9 September: a price is not a cost per candidate until you know how many candidates it buys, and the denominator is published far less often than the price is. Test that here and the result is worse than expected. Of the providers checked today, the one that does publish a price does not publish the credit allowance attached to its own plans on its own pricing page. So the crossover arithmetic this site has printed for three markets — divide the annual plan by your per-person price and see who wins — cannot be computed in this lane from published figures at all.

For a small business in Australia or New Zealand that changes what to do in the first email rather than in the demo. Ask for the allowance in writing before you compare any per-candidate number, including ours. Ask how many candidate sittings the plan includes, what happens in the month you exceed it, and whether unused allowance carries. A vendor who answers those three in writing has given you enough to compare; one who will not has given you a headline price and a demo run at full use, which is the one condition under which every allowance plan looks cheap.

The AssessAll side of that is on the page, which is what makes the question fair to ask: one credit is US$0.50, an aptitude module is five credits, and a five-module screen is therefore US$12.50 a candidate with no plan, no seat licence and no allowance to exceed. A worked specification with every item count and duration behind that figure is published at the national talent development cohort blueprint — a different market, the same arithmetic.